In the first three parts of my series on Email & the UK Election, we have investigated the email deliverability, subscriber engagement, and supporter profiles of each of the main parties contesting the election. Now it’s time to pull these strands together and see whether Return Path’s email intelligence can predict the outcome. This will be our version of fivethirtyeight.com, although I’m not comparing myself with Nate Silver when it comes to our respective statistical abilities! Let me also caveat that my tongue has been firmly in cheek as I’ve prepared this analysis!
Let’s start with current party support as represented by their respective email list sizes:
At face value, the Labour Party is comfortably ahead in terms of support it is enjoying. However, there are a few factors that mean size isn’t everything – let me explain.
Return Path has a nifty piece of functionality in RP Labs that evaluates email lists for the split between primary, secondary, and dead addresses. See below for the Liberal Democrats current address profile:
All programs should have as many subscribers with primary addresses as possible, because they generate greater engagement. Here’s a summary of how all the parties are doing:
The SNP and the Greens have the most engaged lists when measured against this metric. UKIP has a meaningful amount of dead addresses that it would do well to scrub.
Another piece of important intelligence is how much of a list senders share with their competitors. For example, the Conservative party shares 11% of its subscribers with the Labour party!
Extending this analysis across all parties creates the following table:
Subscriber overlap can be a proxy for floating voters. Parties with high levels of overlap (Liberal Democrats, UKIP, Greens) should expect to see much of their notional support evaporating on polling day.
The way supporters engage with their emails may also provide direction about their voting intentions. Positive engagement metrics (Read rates, forwards, not spam notifications, etc.) may reflect positive political engagement, while negative engagement metrics (spam complaints, deleted unread, etc.) may mean the opposite.
All programs are generating Excellent or Above Average engagement benchmarks, although the Conservative, Labour and Liberal Democrat programs are starting to show signs of fatigue, perhaps because of their higher levels of campaign activity.
Email Intelligence Predicts . . .
I’ve used these factors to create a revised version of the original list size breakdown chart, now weighted to reflect address profile, list overlap, and subscriber engagement:
It’s starting to look like a straight two-horse race between the two main parties. However, we must now deal with an anomaly in the UK electoral process, whereby the same level of support but concentrated in a smaller geographical area will deliver more parliamentary seats. For example, the Scottish National Party – with approximately 4% national support but concentrated only in the north of the country – is expected to win over 50 seats. However, the Green party – with a similar level of national support, but spread throughout the UK – is only expected to win 1 seat (if it’s lucky!).
To cater for this, I plugged my percentages into Electoral Calculus which accommodates these biases. This allows me – finally –to arrive at my email intelligence-based prediction!
It’s going to be pretty much a dead heat! The Conservative party will secure most seats (just), but falling short of an overall majority (325 seats), and lacking the legitimacy to form a minority government. This will be the widely predicted hung parliament, and there’s going to be some serious horse-trading over the weekend! So – who will make good stable mates?
Let’s come back to list overlap for a moment. When it comes to coalition building, overlap could be good – it implies commonality of interest. Where the overlaps also show high mutual Read rates, this could be a proxy for willingness to embrace each other’s policies:
This table, showing average Read rates for each overlap segment, calls out high levels of mutual engagement between the Liberal Democrats, and both the Scottish National Party and the Greens (there is also serious antipathy between Conservative and SNP supporters!).
Combining overlap size (representing broadness of shared support) with overlap engagement (representing willingness to work together) produces this final set of indices.
This suggests the Liberal Democrats and UKIP will represent a good fit with Conservative philosophy. However, this would only be a handful of additional seats. Even if all minor parties were also co-opted (unlikely) it still won’t deliver an overall majority for David Cameron.
The Liberal Democrats and the Greens would also be an obvious fit in a Labour coalition, but the same problem of no overall majority exists for Ed Miliband as well.
This leaves Nicola Sturgeon’s Scottish National Party as the king-makers (although that should probably be queen-makers – apologies, Your Majesty!). They are avowed enemies of the Conservatives. And Labour is on record saying there will be no coalition deal with the SNP. But presented with the opportunity to take the reins of power together, I think they could just about find a way to work together.
The same data that we use to drive Return Path’s Email Optimization solutions is predicting the outcome of the 2015 UK general election. Let’s see how close we get!